Implicit conversion from float -10.637450199203187 to int loses precision